How Self-Driving Cars Use Danger Maps to Drive Safer

This patent describes how vehicles can use a 'danger map' of known hazardous locations to plan safer routes and automatically adjust their driving behavior, like speed or following distance, when approaching risky areas.

What it covers

This system helps a vehicle drive more safely by first receiving a 'danger map' (Claim 1). This map shows different locations, each with a 'danger value' based on past events like accidents or near misses collected by other vehicles (Claim 1, 6). The vehicle then uses this map to determine a route from its starting point to its destination, choosing paths with lower danger values (Claim 1). While driving, if the vehicle approaches a location whose danger value exceeds a set limit, it automatically modifies its operation (Claim 2). This modification can include reducing its top speed, increasing its following distance from other cars, or even changing its lane-changing rules (Claim 4, 5). For example, if the danger map indicates a specific highway exit has a high danger value due to frequent emergency braking, a self-driving car might automatically slow down and increase its following distance as it approaches that exit.

What it doesn't cover

• —Does not cover systems that only detect real-time hazards using on-board sensors without relying on a pre-received, location-specific 'danger map' (Claim 1).
• —Does not cover systems that only warn a human driver about hazards without automatically modifying the vehicle's driving behavior (Claim 2).
• —Does not cover systems where the vehicle's operation is modified based on current traffic conditions alone, rather than a pre-determined danger value associated with a specific location (Claim 1).
• —Does not cover systems that only suggest an alternative route to a human driver, without the vehicle itself determining and potentially adjusting its own route based on danger values (Claim 1, 9).
• —Does not cover systems where the 'danger map' is exclusively built by the vehicle's own sensors without receiving data from other sources or vehicles (Claim 1, 6).

The clever bit

The core innovation is combining a pre-existing, location-specific 'danger map' with dynamic, automated adjustments to vehicle operation. This allows a vehicle to anticipate potential hazards at specific points on its route based on historical data, rather than just reacting to immediate sensor inputs, enabling proactive safety measures.

Why it matters

This patent addresses a critical challenge for autonomous vehicles: how to anticipate and react to hazards that aren't immediately visible to on-board sensors. By using shared, location-based danger data, vehicles can proactively adjust their driving, potentially preventing accidents and improving safety. This approach helps autonomous systems move beyond reactive driving to more predictive and cautious navigation, which is essential for public trust and widespread adoption of self-driving technology.

Real-world examples

• 1.Autonomous driving systems (e.g., Waymo, Cruise)
• 2.Advanced Driver-Assistance Systems (ADAS) that integrate map data for predictive speed control
• 3.Navigation apps that factor historical accident data into route suggestions
• 4.High-definition mapping services used by self-driving cars